AI as 'Normal Technology' Means Job Changes Will Take Decades
Narayanan's framework provides a counterpoint to the narrative that AI will rapidly replace human workers, grounding the discussion in historical technology adoption patterns.
Reporting from 1 source: GIGAZINE.
Princeton professor Arvind Narayanan, in a keynote at ICML 2026, argued that AI should be viewed as 'normal technology' like electricity, which takes decades to diffuse. He warned that the idea that productivity gains from AI will lead to mass unemployment contradicts economic data, citing the Jevons paradox. Instead, job roles will slowly transform, requiring organizational adaptation over many years.
Arvind Narayanan delivered the keynote at the International Conference on Machine Learning in Seoul on July 6. His talk, titled 'What challenges should we tackle?', laid out a framework for understanding AI's economic impact. He argued that AI is 'normal technology'-like electricity or the internet-that takes decades to permeate society. The diffusion stage, especially adaptation, is the slowest and has not yet begun in earnest. Narayanan also pointed to the Jevons paradox, where increased efficiency often leads to greater overall resource use, not less. He said the idea that a tenfold productivity gain would mean one-tenth the workforce contradicts actual data.
Synthesized by Yomimono from the 1 cited source below, including Japanese-language reporting where cited, then editorially reviewed before publishing.